c14kevincardenas/mobilevit-small_alpha0.5_temp3.0_t2

mobilevit-small_alpha0.5_temp3.0

This model is a fine-tuned version of c14kevincardenas/ClimBEiTv2 on the c14kevincardenas/beta_caller_284_person_crop_seq_withlimb_2 dataset. It achieves the following results on the evaluation set:

• Loss: 0.8337
• Accuracy: 0.6789

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-05
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.2251	1.0	90	1.4083	0.2441
1.1754	2.0	180	1.3374	0.3063
1.086	3.0	270	1.2295	0.3844
0.9164	4.0	360	1.0471	0.5208
0.8124	5.0	450	0.9794	0.5692
0.757	6.0	540	0.9078	0.6136
0.7021	7.0	630	0.8754	0.6423
0.6535	8.0	720	0.8976	0.6206
0.5973	9.0	810	0.8455	0.6719
0.5778	10.0	900	0.8477	0.6532
0.5556	11.0	990	0.8395	0.6660
0.5226	12.0	1080	0.8487	0.6542
0.4926	13.0	1170	0.8610	0.6482
0.4766	14.0	1260	0.8337	0.6789
0.4763	15.0	1350	0.8482	0.6650
0.45	16.0	1440	0.8374	0.6700
0.4468	17.0	1530	0.8378	0.6759
0.4267	18.0	1620	0.8382	0.6729
0.4321	19.0	1710	0.8399	0.6719
0.4318	20.0	1800	0.8438	0.6729

Framework versions

• Transformers 4.45.2
• Pytorch 2.5.0+cu124
• Datasets 3.0.1
• Tokenizers 0.20.1